This proposal seeks to improve the accuracy of noninvasive diagnosis and prognosis of coronary artery disease by using MRI and a gadolinium-based paramagnetic contrast agent. First-pass MRI with the modeling methods proposed here may be able to provide absolute regional blood flows at a high spatial resolution. These first-pass studies may also offer unique viability information. The specific aims are (1) To develop and optimize acquisition strategies to obtain data tailored for compartmental modeling (2) To develop clinically practical methods for analyzing the cardiac contrast MRI data with models (3) To compare the flow estimates from the MRI methods developed here to an MRI upslope perfusion index and to absolute blood flow measurements obtained with dynamic N-13-ammonia PET. (4) To add viability measures from the first pass modeling approach to delayed images to determine if such an approach improves prediction of viability. Methods: (1) Systematic analysis of temporal sampling strategies and the use of reduced k-space acquisitions to increase volume coverage, reduce artifacts, and maintain signal and high spatial resolution will be pursued using realistic computer simulations and human studies. (2) Linked active contours combined with temporal clustering methods will be developed to automatically segment the endocardium and epicardium in the time series data. Methods for blind identification of the input function (input functions are inaccurate at high gadolinium concentrations) will be developed and validated. Two different physiological models will be developed and compared in their ability to provide absolute flow values and reliable extracellular volume estimates. (3) The comparisons with MRI upslope and PET perfusion will be performed using 34 human studies. (4) The first-pass viability measures and three integrated viability measures will be compared to delayed enhancement Gd MRI images and to post-revascularization data in 17 patients. The outcome of this project will be validated imaging protocols and software for use with first-passMRI studies, and practical and accurate methods for myocardial perfusion and viability assessment in vivo. Such methods will be invaluable for improved health care and for improved basic science research, such as tracking nascent flow changes in gene therapy.